While Red Hat Enterprise Linux 5 will be released at an event in San Francisco on March 14, the company has not been sitting idle since the code went gold last month, with its engineering team already doing future planning and hosting worldwide summits, the first of which was held in February.
Red Hat is also developing a new model for next-generation capabilities that allows it to continue to work with the open-source community while also developing early service-based relationships with customers around these new technologies.
The goal of this is to let customers help influence the direction of next-generation technologies and capabilities, while putting that code into production before it is widely released, Brian Stevens, Red Hats chief technology officer and vice president of engineering, told eWEEK in a recent interview.
This will also allow Red Hats engineers to work more closely with those customers to understand their environments better. “This service relationship will be unique and unlike anything we have done in the past in terms of how upgrades work and sustainability,” he said.
“So those customers are really opting into a different set of rules, and we are designing those rules around what makes sense. One of the considerations we are having specifically with the new real-time technology is how we harden it while allowing it to advance,” he said.
As such, Red Hat has not developed a pricing model for this yet, but it will be a service-based relationship to a limited set of customers and will be used for both its new messaging technologies and the real-time Linux development work currently under development. “But they will both still be subscription-based relationships,” Stevens said.
The new messaging technologies and the real-time Linux development work are truly differentiating and will have impact, creating real value for customers, so cost is really the footnote in the conversation between Red Hat and its customers, who want to know how these can be used to create value for them, he said.
Growing Red Hats business while creating value for its customers are overlapping goals, Stevens said.
“In the case of the real-time work, its an alternate kernel so we are not talking about capability thats fully merged upstream. So its not just a layered product or technology set, its a support arrangement around an alternate patch set that we are managing and which we will ultimately drive mainstream so we will have a unified product down the road that will be made generally available,” he said.
New in RHEL 6
New in RHEL 6
Asked by eWEEK what new technologies will form the basis of RHEL 6, the next version, and beyond, Stevens said the company, headquartered in Raleigh, N.C., has ramped up its team working on scalable message queuing, which is not part of RHEL 5, and is creating a community of developers under the Apache umbrella, known as Qpid as well as a specification working group, AMQP.
“We are going to be bringing that, under subscription, to a small number of customers over the next quarter, starting with a lot of financial service firms who will be using it for market data management and the like,” Stevens said.
“But, once we make that broadly available, we expect to build policy-based management on top of the framework. So its not just around critical market data, but around how you build an agile data center that is event-driven in a scalable fashion and away from the legacy systems management monitoring products of the past,” he said.
As such, Red Hat is going to “plumb into this messaging model for next-generation data center management, which will be a big theme over the next release,” Stevens said.
There will also be a continuation of the virtualization technology, the first phase of which is found in RHEL 5, as well as delivering on policy-based management and trying to automate that data center and grid.
Red Hat is also looking at the integration of high-availability capabilities with these virtualization capabilities, so as to obviate the need for things like Oracle RACs (Real Application Clusters), which Stevens said are “insanely priced” and of limited value.
“When you start to look at frameworks of high availability, of RHEL 5 integrated with Xen, with online migration and high availability for any database type, and combined with multicore, the scale you hit is just insane, and there is no reason to go for horizontal scaling any longer. So there will be work done on all that,” Stevens said.
In addition, Red Hat will be expanding on the notion of a database container, and taking RHEL 5 and creating database containers that let IT administrators do all the things they could previously do only with expensive RAC environments with PostGRES, Sybase, Ingres and others, he said.
Stevens also gave his views on why competitor Oracle—which is challenging Red Hat with Oracle Linux—has not embraced virtualization, while considering the threat posed to Oracles RAC business now that database containers coan be built using virtualization.
“Virtualization was disruptive to Oracles business, and they knew it, and now it has arrived. Any database can now be put into a virtual machine, fully managed, with both automatic and manual resource assignment, with memory and CPU being able to be added and removed on the fly,” he said.
“So, what we are talking about is a common infrastructure, a common platform for applications that all need these services. So now, all of a sudden, we can bring high availability and all of these capabilities, not just to an Oracle RAC environment, but to any database. So we have totally leveled the playing field between databases through the notion of this database container,” Stevens said.
This, said Stevens, means that RAC, with a 50 percent uplift per CPU, is now obviated.
“The scalability is not needed and it never really worked because of all the multicore advancements, and the high availability is now inferior technology as IT professionals do not want to invest in a high-availability strategy per application. Im looking forward to what the future brings,” he said.